Hi Michael,
I guess the answer for the reviewer would be that it does not make sense to normalise the data twice, which would be in fact what you would do if you were to include the TIV in an FSL-VBM analysis (as you have already normalised the images when you affine registered them to a common template and did *not* correct afterwards for such effect with the modulation)...
If a reference is needed, then Good et al., 2002 (not the 2001):
"Since we have incorporated this correction for volume
change induced by spatial normalization, it is appropriate
to include TIV as a confounding covariate to
remove any variance due to differences in head size.
Without the modulation step, spatial normalization
removes differences in head size and thus TIV does not
need to be included as a confounding covariate."
In this paper, the modulation step (as opposed to FSL-VBM) takes into account the affine part, therefore, with it, you do correct for removing differences (I know, it's hard to follow). In other words, by modulating the data in the Good's study, they did reintroduce differences in head size, hence the TIC as a covariate. *Not* with FSL-VBM.
Hope it makes sense,
Gwenaelle
--- En date de : Jeu 10.12.09, Michael Hornberger <[log in to unmask]> a écrit :
> De: Michael Hornberger <[log in to unmask]>
> Objet: Re: [FSL] Re : [FSL] VBM
> À: [log in to unmask]
> Date: Jeudi 10 Décembre 2009, 21h46
> Hi all,
>
> Thanks for this info Gwenaelle.
> The issue of TIV correction seems to come up quite often
> with reviewers.
> Is there a paper one can cite which shows that TIV
> correction is not
> necessary if the modulation does not include affine
> registration?
>
> Thanks,
> Michael
>
>
>
> Gwenaëlle DOUAUD wrote:
> > Hi Gabor,
> >
> > you don't need any correction for TIV. The Jacobian
> modulation does *not* include the affine part of the
> registration, which means that your data are already
> normalised for such scaling effect. You need to correct for
> TIV if the modulation includes the affine part.
> >
> > Anyway, you can always calculate the TIV by adding the
> volumes of the 3 pve images (fslstats -M x fslstats -V)
> obtained after the fast segmentation.
> >
> > Cheers,
> > Gwenaelle
> >
> >
> > --- En date de : Jeu 10.12.09, Gabor Perlaki <[log in to unmask]>
> a écrit :
> >
> >
> >> De: Gabor Perlaki <[log in to unmask]>
> >> Objet: [FSL] VBM
> >> À: [log in to unmask]
> >> Date: Jeudi 10 Décembre 2009, 10h31
> >> Hi!
> >>
> >> Is it possible to correct for total intracranial
> volume in
> >> FSL?
> >>
> >> Thanks, Gabor
> >>
> >>
> >
> >
> >
> >
>
